Air Transportation is a Complex
Adaptive System:
Not an Air Traffic Control
Automation Problem
Dr. George L. Donohue
George Mason University
March 18, 2004 Harvard University
© George Donohue 2004
Outline
• How did We Get Here?
• Why Should We Care?
• Capacity - Delay
• Capacity – Delay – Safety
• System Network Effects
• Observations and Recommendations
How did We Get Here?
• 1903 Wright Bros. produced a Heavier than Air Flying
System: AGE OF INVENTION
– Airfoils, L/W Structure, Controls, L/W Propulsion
• WW II ( +50 Yrs) System is Upgraded: AGE OF ALL
WEATHER COMMERCIAL FLIGHT
– Radar, Jet Aircraft, Radio Navigation and Communication
• 2003 ( +100 Yrs.) System Needs Upgrading Again: AGE OF
RELIABLE INTERNATIONAL TRANSPORTATION
NETWORK
–
–
–
–
Predictable under all Weather Conditions
Maximum Airport Capacity Utilization
Near Optimal Network Load Balancing
Predicable Safety Operating Margins
Barriers to the Third Age Vision
• The Technical Community has been Aware of the
Transition Problem for Over 20 Years!
– Increasing Delays, Flight Cancellations
– Increasing Runway Incursions, ATC Op Errors and TCAS RA’s
• The Technical Elements that will enable the Development of an
Affordable, Reliable, and Predictable Mode of International
Transportation Already Exist!
–
–
–
–
–
TCAS II Deployed Worldwide
FMS with ±30 Sec. RTA ?
GPS Navigation and Surveillance (ADS-B)
Digital Communication Data Links
TMA, pFAST, aFAST, URET
No System Credit
No System Credit
No System Credit
NOT DEPLOYED
No System Credit
• The Barriers to Growth are Regulatory and
Institutional!
Increasing Delay is a Frog in the
Boiling Water Problem
Growth rate/GDP growth rate
4.50
4.00
3.50
3.00
Deregulation
Air carrier
passenger miles
2.50
Highway trip miles
2.00
Rail passenger miles
1.50
1.00
0.50
0.00
1960
1965
1970
1975
1980
1985
1990
1995
1998
Year
Source: USDOT BTS NTS 2000; USDOT BTS update April, 2002; DOC BEA 2002 (*real GDP using 1996 chained dollars
Air Transportation’s
Contribution to GDP
E co n o m ic Im p act o f Aviatio n In d u stry
($b illio n s 1999)
T o tal Ou tp u t
Air Transportation
Aircraft Manufacturing
Tourism
Agents/Forwarders
Government
T o tal Im p act
$205
$134
$94
$3
$2
$438
G DP Co n trib u tio n
$80
$94
$85
N/C
N/C
$259
N/C = Not calculated
Source: L. Anderson, presentation to Aeronautics and Space Engineering
Board, June 1999
FUTUREMCivil
ARKETS
FOR
Transport
Share of
AERONAUTICSPRODUCTS
ARELARGE
Aerospace
Industry
Total Projected Aircraft Market 1999 to 2008: $810 Billion
Regional /
Commuter
Business /
Corporate
Military
General
Transport
Aviation
$31
$6
74% CIVIL
TRANSPORT
$54
$65
Rotorcraft
(Military and $81
Civil)
Large Civil
Transports
$473
Fighter, Attack
and Trainer
$100
Capacity and Delay
• System Capacity is Primarily Limited by Network
Runway Availability
• ATC Workload is an important Secondary
Limitation
• Runway Maximum Capacity is a function of
Aircraft Landing Speed and Runway Occupancy
Time (ROT)
• Delay is a Non-Linear function of Demand to
Maximum Capacity Ratio
– Stochastic FCFS System
– Queuing Theory Applies
• Major Hub Airports are Over-Scheduled
– Transportation Network Is NOT LOAD BALANCED
– Market Mechanisms Could Achieve this Goal
Network Operational Capacity is
a Limited Commodity
•
C
MAX
C
S i (XG)i Ri
{Airports}
–K AK(t) {Airspace Management Intervention}
=2
AR MAX
• S = f ( Safety, ATC , Wake Vortex, etc.) ~ 0.6 to 0.8
• AK(t) = (A/C
– A/CACCEPT) ~ [ 0 to >1,000]
– AK(t) = f ( GDP:Weather, Sector Workload Constraints )
REQUEST
• C
~ 40 Arrivals/Hour (set by Runway Occupancy Time)
• Ri = Number of Runways at ith Airport
• XGi = Airport Configuration Factor at ith Airport
AR MAX
• i = 1 to N, where N is approximately 60 Airports
• K = 1 to M, where M is typically much less than 100 Sectors
ATS Delays Grow Exponentially
with Increasing Capacity Fraction
NUMBER DELAYS
( >15 min / 1000 operations)
PREDICTED DELAY vs. CAPACITY FRACTION
50.0
45.0
40.0
35.0
30.0
25.0
20.0
15.0
10.0
5.0
0.0
0%
20%
40%
60%
80%
AIRPORT CAPACITY FRACTION
100%
Aircraft Arrival Rate:
Distance-Time Relationship
Spacing
80
(sec)
ARRIVALS / RW / HR
70
120 Knots
60
60
130 Knots
ROT?
50
72
140 Knots
90
40
WV?
30
120
180
20
10
0
0
1
2
3
4
DISTANCE ( NMi)
5
6
7
NY LaGuardia: A non-Hub
Maximum Capacity Airport
• 1 Arrival Runway
• 1 Departure Runway
• 45 Arrivals/Hr (Max)
• 80 Seconds Between Arrivals
• 11.3 minute Average Delay
• 77 Delays/1000 Operations
• 40 min./Delay
New York LaGuardia Airport ArrivalDeparture Spacing VMC
60
ASPM - Apr 2000 - Visual Approaches
ASPM - Oct 2000 - Visual Approaches
Calculated VMC Capacity
50
Arrivals per Hour
Optimum Rate (LGA)
40
40,40
Each dot represents one
hour of actual traffic
during April or October
2000
30
45 Arr./Hr/RW
@ 80 sec
separation
20
10
DoT/FAA
0
0
10
20
30
40
Departures per Hour
50
60
LGA Arrival - Departure IMC
60
ASPM - April 2000 - Instrument Approaches
ASPM - October 2000 - Instrument Approaches
Calculated IMC Capacity
50
Arrivals per Hour
Reduced Rate (LGA)
40
32,32
30
20
10
0
0
10
20
30
40
Departures per Hour
50
60
Capacity-Delay-Safety
• ATM System Safety and Capacity are NonLinearly Related
• Wake Vortex Separation sets the Current
System Capacity Limit in Instrument
Meteorological Conditions
– Safety Limitation
• ICAO System Safety Goal is 10-9 / Operation
• Small number Statistics leads us to use Accident
Precursors as Safety Indicators
• Safety Analysis must be Analytical
Time Separation Is an Important
Determinant of the SAFETY Limitation
35+ Arrivals/RW/Hr
ROT
Probability
A/C Inter-arrival Time
Time (seconds)
Accident Pre-Cursor Incidents
Safety – Capacity Relationship
Hazard Reports 1988-2001
120
NMAC
Number Reports Filed
100
RWY Inc
80
Legal Sep
60
40
20
0
35
40
45
50
55
60
65
70
Percentage Capacity Used
ATL, BWI, LGA, DCA
Haynie, GMU 2002
ATL Estimated
Collision Probability
Collision probability per SRO for each combination
Small-Heavy
Leader - Trailer
Small-Large
Small-B757
Richard Y. Xie, GMU research in progress
Wake Vortex Accident Rate in
Safety-Capacity Coordinates
Single Runway Estimated Wake Vortex Accident Rate
50% Mix B747 & B737: S-Wake Calculation
450,000
Log. (Hazardous Accident)
Capacity - Arrivals / Year
400,000
Log. (Catastrophic Accident)
350,000
300,000
250,000
200,000
150,000
100,000
y = -37168Ln(x) + 832913
R2 = 0.9698
30 Years
y = -25004Ln(x) + 593477
R2 = 0.8731
3 Years
50,000
-
1,000,000
2,000,000
3,000,000
4,000,000
5,000,000
Safety - Arrivals / WV Accident
NLR Stochastic Analysis
6,000,000
7,000,000
System Network Effects
• Aprox. 10 Major Hub Airports are
Operating at D/C max > 0.65
• Delays at these Airports spread NonLinearly throughout the Network
• Runway Additions at one Airport May have
Little Network Effect
• System-wide improvements have a Larger Effect
than Individual Airport Improvements
– Except at Major Airports like ORD, LGA and ATL!
Major US Airport Congestion
LAX
ATL
STL
ORD
SEA
MSP
AIRPORT
LGA
SFO
PHL
EWR
IAD
DTW
Queuing Delays
Grow Rapidly
DFW
CLT
PIT
J. D. Welch and R.T.
Lloyd, ATM 2001
JFK
BWI
DEN
0
0.1
0.2
0.3
0.4
0.5
0.6
DEMAND / CAPACITY RATIO
0.7
0.8
0.9
Market Does Not Act to Minimize Delay or
Maintain SAFETY: LGA Air 21 Impact
LaGuardia Airport
200
180
160
140
120
100
80
60
40
20
0
Maximum Hourly
Operations Based on
Current Airspace &
ATC Design
06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Time of Day
Historic Movements
AIR-21 Induced Svc.
Source: William DeCota, Port Authority of New York
Atlanta: A Maximum Capacity
Fortress Hub Airport
•
2 Runways – Arrivals
•
2 Runways – Departures
•
50 Arrivals/Hr/RW – Max
•
72 Seconds Between Arrivals
•
8.5 minutes Average Delay
•
36 Delays/1000 Operations
•
38 min./delay
Simulated Auction
Delay Benefit at ATL
Scheduled arrivals (#operations/quarter hour)
50
40
30
ATL reported
optimum rate
20
10
0
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Estimated Average Runway Queuing Delay (min)
20
15
10
5
0
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
Time (15-min bins)
Original Schedule
Loan Le Research
in Progress
Auctioned Schedule
45 min maximum schedule deviation allowed,
no flights are rerouted
Observations – NAS Safety
• We are approaching the Point that the
existing system may be demonstrably less
safe (at current and future capacity
fractions) than a new, more synchronous,
aircraft FMS/ADS-B separation based
system
• System is Safe BUT Safety Margins are
Diminishing!
• This case has not been Analyzed nor even
Suggested as a RATIONAL for CHANGE
to date!
Central Research Questions
• Both Safety and Efficiency Concerns lead us to the
conclusion that the network should be operated as a
Synchronous System
–
with economic incentives to use the largest aircraft
affordable and economically viable
• Time Window Auctions at Airport Metering Fix
may provide the Economic Incentives necessary to
maximize/optimize Network Capacity
• Central Research Questions:
– How Synchronous Can We Make this
System – All Weather?
– What Should the Design Target Level of
Safety Be?
• Backup Slides
10
8
6
4
2
0
140
100
60
20
-20
VFR 33.8 Arr/hr
IFR 34 Arr/Hr
VFR 30.9 Arr/Hr
VFR 27 Arr/Hr
-60
Aircraft / RW / Hr (20 Sec.
Bins)
Wake Normalized Aircraft Time
Separation: LGA in VMC & IMC
Seconds Deviation per Aircraft From
Perfect WVSS Adherence Value
Observed WV Separation
Violations vs. Capacity Ratio
Number of < WVSS
Incidents Expected in
15 Minutes
Figure 6-5
Ratio of Incidents to Capacity Used
8
6
4
2
0
0
50
100
150
200
Percent of Capacity Used in 15 Minutes
BWI
LGA
Quadratic Model
Haynie, GMU 2002
ATL Arrival - Departure IMC
120
ASPM - April 2000 - Instrument Approaches
Calculated IMC Capacity
84,90
100
Arrivals per Hour
Reduced Rate (ATL)
80
60
40
20
0
0
20
40
60
80
Departures per Hour
100
120
ATL and LGA Inter-Arrival Time in IMC
and VMC:32 - 39 Ar/Rw/Hr
LGA & ATL Arrival Histograms
14
LGA in VMC N=168
Aircraft / RW / Hr (20 Sec. Bins)
12
LGA in IMC N=124
10
ATL IN VMC N=114
8
ATL in VMC N=323
6
4
2
0
0
50
100
150
-2
Inter-Arrival Time (Seconds)
200
250
Observations - Recommendations
• FAA Culture – Barriers to Change
• NASA Culture – Barriers to Change
• State of NAS Safety
• Proposed Grand Experiment/OPEVAL
FAA Barriers to Change
• FAA has an Operational and Regulatory
Culture
– Inclination to follow training that has seemed
to be Safe in the Past
• FAR has NOT Changed to Provide
Operational Benefits from Introduction of
New Technology
• Assumption that Aircraft Equipage would
be Benefits Driven did not account for
Lack of an ECONOMIC and/or SAFETY
Bootstrapping Requirement
FAA Investment Analysis Primarily
focus on Capacity and Delay
• OMB requirement to have a B/C ratio > 1
leads to a modernization emphasis on
Decreasing Delay
• In an Asynchronous Transportation
Network operating near it’s capacity
margin, Delay is Inevitable
• Delay Costs Airlines Money and is an
Annoyance to Passengers BUT
– is Usually Politically and Socially Acceptable
NASA Barriers to Change
• NASA has become more Process Oriented than
Product Oriented
• Frequently Stated Objective of 25 year
Implementation Goal Avoids Accountability and
renders NASA TRL 4/6 Product unusable by either
Government Agencies or Industry
• NASA needs a Cadre of Engineers/System Analysts
with a long range goal of becoming the USG
Technical Experts in Aviation System Safety
Analysis
Proposed Grand Experiment:
OPEVAL to FOCUS Efforts
• FY 2008 One Year of Night Operations
– 12pm to 8 am
• DAG-TM + aFAST+CDM + WV
• Entire US Air Cargo Fleet
• Inter-Agency IPT
– DoT, NASA, FAA, DoD, NTSB, Boeing, CAA,
Airlines
Hypothesis: Most Major Changes to the
NAS have been due to Safety Concerns
• 1960’s Mandated Introduction of Radar Separation
• 1970’s Decrease in Oceanic Separation Standards
Required a Landmark Safety Analysis
• 1970’s Required A/C Transponder Equipage
• 1970’s Required A/C Ground Proximity Equipage
• 1990’s Required A/C TCAS Equipage
• 1990’s Required A/C Enhanced Ground Prox.
Equipage
• 1990’s TDWR & ITWS Introduction
• 1990’s Mandated Development of GPS/WAAS